Available for remote work

Research-oriented engineer · Manizales, Colombia

Engineering systems that make their evidence visible.

I’m Martín Ramírez Espinosa—an Electronic Engineering and Mathematics student building inspectable research and machine-learning systems, local-first products, and reliable software.

Available for remote internships, part-time engineering roles, and focused contracts alongside my studies in Colombia.

Portrait of Martín Ramírez Espinosa

Selected work

View all six projects
RAGdoll research with receipts project banner

Research systems · AI

RAGdoll

Scholarly research, with receipts.

A terminal workspace that turns ambiguous research questions into curated, cited literature dossiers while preserving human approval, retrieval provenance, evidence level, and exact supporting passages.

  • Python
  • Textual
  • SQLite FTS5
  • OpenAI
  • Ollama
OpenAI Build Week submission · v2.2 release candidate
Sealed-test bits per character results for smaLLM tokenizers

Language models · Reproducible research

smaLLM

A GPT-style language-model lab built from scratch.

An inspectable PyTorch implementation spanning tokenization, causal attention, training, generation, controlled baselines, sealed evaluation, and reproducible experiment artifacts.

  • Python
  • PyTorch
  • Transformers
  • Experiment design
Completed research artifact
insIGht local Instagram export import interface

Desktop product · Privacy

insIGht

Instagram relationship analytics without an Instagram login.

A local-first desktop application that turns official Instagram exports into searchable relationship snapshots without scraping, telemetry, cloud processing, or unofficial APIs.

  • Rust
  • Tauri
  • React
  • TypeScript
  • SQLite
Cross-platform desktop application

Technical notes

The reasoning behind the repository.

View all six notes

RAGdoll · Architecture note

Architecture and trust boundaries

How provider, scholarly-source, storage, evidence, and terminal boundaries preserve user control and keep model output outside domain logic until validation succeeds.

  • System architecture
  • Trust boundaries
  • Pydantic
Read the artifact

smaLLM · Experiment report

Final capacity-controlled tokenizer panel

The preregistered final comparison of boundary-aware ByteBPE512 and matched character controls, including every seed, corpus, checkpoint identity, reversal, and limitation.

  • Experiment design
  • Tokenization
  • Sealed evaluation
Read the artifact

One engineering method

Different domains. The same standard of evidence.

01

Make boundaries explicit

Keep models, parsers, storage, hardware, and user approval behind contracts that can be inspected.

02

Measure before claiming

Prefer controlled comparisons, reproducible artifacts, and documented negative results over impressive anecdotes.

03

Design for user control

Keep sensitive data local, permissions visible, and failure states honest instead of hiding uncertainty.

What comes next

Looking for work where research rigor meets real systems.

I’m available for remote internships, part-time engineering roles, and focused contracts alongside my Electronic Engineering and Mathematics studies in Colombia.